Skip to content

Choose a starter setup

Begin with one real task and a way to check its output. Coding experience is not required to follow the course; the GUI-only walkthrough is still to be prepared.

Working notes

  • Write down the question, expected artifact and stopping condition.
  • Keep a brief and results log outside the conversation.
  • Choose an environment that fits the experiment’s lifetime.

Quick starts

  1. Create a disposable working folder with a short task brief.
  2. Open your chosen agent interface and give it a small, checkable task.
  3. Inspect the output before expanding the scope.
  4. [TODO: add tested installation steps and a zero-terminal browser example.]
  1. Put the brief, code and results under durable versioned storage.
  2. Separate work into sessions and worktrees with bounded responsibilities.
  3. Check progress through artifacts and heartbeats.
  4. [TODO: add a tested persistent-session recipe and recovery exercise.]
  1. Read local login-node, queue and resource policies.
  2. Run a small pilot in a scheduler allocation before a full study.
  3. Record job IDs, event counts and completed artifacts outside the session.
  4. [TODO: add a site-approved submission example and compute budget.]

Start small

Bring a question and a known reference result. The volunteer setup will be built around these, not around a tool shopping list.

Before the course

  • [TODO: select a compact worked example and its verification artifact.]
  • [TODO: rehearse the live walkthrough and prepare a fallback.]

Back to the deck ↩